Senior Scientist, Dewpoint Therapeutics
|Type||Kitchen Table Talk|
D. Allan Drummond, associate professor at the University of Chicago, joined the Dewpoint scientists and Condensates.com community for a fantastic seminar on November 2, as part of the Kitchen Table Talk series. Allan has been a professor since 2011, when he started his academic research lab as a Bauer Fellow at Harvard after completing his PhD at Caltech with Nobel Laureate Frances Arnold, where he studied protein evolution.
Interestingly, Allan spent seven years at the prominent software company Trilogy, ultimately culminating as the director of HR where he had a real emphasis on mentorship, which he continues to foster in his own lab. There, Allan also continues to meld his knowledge of computation and protein evolution to ask and answer questions about organismal fitness with a variety of multidisciplinary techniques, from hardcore biophysical techniques to genetics. He gave an engaging talk about his lab’s work (see below), and we had an insightful discussion afterward. He was also kind enough to answer a lingering question via email; you can find that further below.
Create an Account or Sign In to view the video.
Bede Portz (00:00:01):
Hi, everybody. It’s a real pleasure to introduce Allan, who’s an associate professor at the University of Chicago in the department of Biochemistry and Molecular Biology where he has been since 2011. Prior to this, Allan was a Bauer Fellow at Harvard, where he studied for five years after completing his Ph.D. at Caltech with Nobel Laureate, Frances Arnold. Allan’s current research program is at once a departure from and an extension of his work prior to the establishment of his own group. And by that I mean previously, he used strictly computational methods, but now he has a wet lab. But some of the questions converge on evolution, which is sort of an ongoing theme of Allan’s work.
Bede Portz (00:00:44):
Specifically to condensates, what I feel has distinguished Allan’s group are two things. One, the multidisciplinary nature of his work. So, he has papers that, in single papers the methods span from hardcore biophysical techniques to genetics. And another emphasis, and I think distinguishing element of Allan’s research program, is an emphasis on function and really organismal fitness, which is perhaps the highest bar sort of measuring function.
Bede Portz (00:01:20):
Two more things about Allan that are interesting: Science was actually not his first career; he spent seven years at, what at the time was, a very prominent software company Trilogy, ultimately culminating as the director of HR where he had a real emphasis on mentorship. And that’s something that continues to this day. In fact, a number of people have left Allan’s lab and have gone on to start their own independent labs, and a number of folks have left his lab and are thriving in postdoctoral roles and elsewhere.
Bede Portz (00:01:53):
Not only is this not Allan’s first career, it’s also not his only career now. He’s also a very accomplished artist, particularly a sculptor, and he has exhibited these incredibly detailed sculptures of insects, both extant and extinct, in galleries. So, Allan is really a marvelous and well-rounded person. I really look forward to his talk today. If it has any of the enthusiasm and sort of curiosity that he’s exhibited in our conversations over the years, I’m sure his talk’s going to be excellent. So, welcome, Allan. Thanks for being here.
D. Allan Drummond (00:02:30):
Well, thank you so much for that, Bede. That is incredibly kind, and everyone should know that I’ve known Bede for many years now. We’ve been actually in each other’s orbits for many years now. And from the minute I met him, there was this sort of instant likability and connection and also scientific respect that has continued to develop over the years. So, we are constantly going back and forth on DMs on Twitter, and it’s lovely to finally see you quasi in person. Thanks, Jill, and thanks to all of you for the invitation here. This should be fun. I hope I can live up to the introduction.
D. Allan Drummond (00:03:11):
The idea here is going to be to try to convince you that one of the most well-studied “stress responses” across the entire tree of life, but, of course, many of us know it mostly from eukaryotes, has really been looked at through a lens that is not quite as productive as it could be, and that once you recognize that what is happening to cells during these sort of stress responses involves condensation, you start to reframe the entire problem in a way that really opens up a whole new set of vistas, I think, are biologically important, that can be inspiring for how we think about biological systems, and also potentially can change the way that we try to manipulate, perturb these systems…
D. Allan Drummond (00:04:04):
It’s important for me to say, I’m a fundamental scientist, a basic scientist, and so almost nothing I’m going to say to you today is going to be about the systems probably that you’re most familiar with if you’re working on mammalian cells, if you’re working on therapeutic approaches to perturbing condensates in mammalian cells. And yet, I promise you, if you try to look through what I’m saying, it will not be a waste of your time because of the loveliness of conservation across the tree of life. Almost everything we’re going to be talking about is shared in virtually every eukaryotic cell that we’ve studied. And by we, I mean humanity.
D. Allan Drummond (00:04:40):
Okay, so the heat shock response as a condensation response, ultimately I’ll say, is a condensation response. So, first, before I dive in, I want to thank my lab. We are small, but we are mighty. Currently, there are these eight of us underneath the fall foliage at the University of Chicago, couple of undergrads, one postdoc, and the rest graduate students. And I’m also going to talk a little bit about the work of Edward Wallace, who’s now the equivalent of tenured at the University of Edinburgh, and Josh Riback, who’s pictured down here. Can folks see my pointer just to make sure? Great. Okay. Josh has just started his group at Baylor College of Medicine as an assistant professor. And Haneul Yoo has just joined Ibrahim Cissé as a postdoc at the MPIIE in Freiburg.
D. Allan Drummond (00:05:36):
All right, so you probably know, but I’d like to remind you, that many stresses trigger similar intracellular changes. Imagine there’s a cell. This happens to be a yeast cell, but we can put any other cell type here. Stress hits the cell, and they respond in very, very stereotypical ways. There is a gene expression response that is induced to respond to the stress. Oftentimes, it includes molecular chaperones, proteins that help other proteins to fold. These stress response messages are translated specifically by a, as yet, unknown mechanism. Most translation of the cells shut off, and a bunch of clumps form. Of course, I don’t need to tell you what these clumps are, but we’ll get there.
D. Allan Drummond (00:06:28):
This happens in heat shock. It happens in nutrient depletion. It happens in hypoxia, heavy metal stress. This super wide range of stresses, which should, I hope in your mind, automatically get you guessing that there’s probably not some simple biophysical trigger that is involved here just because of the fact that like heat and hypoxia are in some sense utterly different things. Of course, they’re going to converge probably in some consistent pathway, but there’s no sense in which we sort of expect exactly the same biophysics to be playing out here. It’s a cellular response.
D. Allan Drummond (00:07:09):
Now, my lab effectively just says, “Okay, well what’s going on here, and how are these responses connected?” We study heat shock, which is sort of the primordial stress of this type. If you heat shock cells, stuff clumps up. They can use molecular chaperones, and the stuff stops being clumped up. That’s the observation from the heat shock field since virtually since its discovery, the discovery of the response in the 60s.
D. Allan Drummond (00:07:37):
This happens, as I mentioned, all over the tree of life, and one of the ways that this has been most commonly studied is using microscopic techniques to look at things called stress granules. Here are a bunch of examples. You don’t need to gaze at them. I will just tell you what these things are showing are specific proteins and poly(A) RNA that in normal, supposedly happy cells–although these are HeLa cells–are dispersed throughout the cytoplasm. And then in response to heat shock and a range of other stresses, clump up into these obvious foci, clusters, that are called stress granules. They happen in human cells. They happen in Drosophila cells. They happen in yeast cells. Literally every cell type where we’ve looked for these things, we find them.
D. Allan Drummond (00:08:29):
What’s interesting about them is we have no idea what they are really or what they do. We just published a review earlier this year in Molecular Cell, talking about the grand challenges, and one of them is that if you want to take an activity-based or a function-based or a fitness-based approach to studying stress granules, you can’t. Not yet. We just don’t know how to interpret the activity of these things. Nonetheless, they fall into this category of, well, you stress things out and a bunch of stuff aggregates, which is the field that I grew up in, the field of protein aggregation, protein quality control.
D. Allan Drummond (00:09:05):
And the heat shock response itself, everybody knows what’s happening in this particular case: Temperature hits a cell, they turn on heat shock proteins using a transcription factor called HSF1, shut down translation, and in many cases the translation initiation factors are the components of these stress granules, so they’re taken out of solution where they’re presumably active and slammed together in inactive particles. These aggregates form, and we know what the aggregates are, “know”.
D. Allan Drummond (00:09:44):
So, I just love this quote from Barbara McClintock’s Nobel essay in 1984, where she lays out a bunch of things that I think we lost track of along the way. She writes, “There are shocks that a genome must face repeatedly and for which it is prepared to respond in a programmed manner. Examples are the heat shock responses in eukaryotic organisms and the SOS responses in bacteria. Each of these initiates a highly programmed sequence of events within the cell that serves to cushion the effects of the shock. Some sensing mechanism must be present in these instances to alert the cell to imminent danger and to set in motion the orderly sequence of events that will mitigate this danger.”
D. Allan Drummond (00:10:24):
I tried to emphasize a few words there: sensing mechanism, imminent danger, cushion. But this is not what was thought to happen during heat shock, actually, in the field. Instead, this is what is thought to happen. Now, there’s a lot going on in the slide. You really don’t need to pay attention to anything here except for the fact that this … And this is taken from somebody else’s review, a prominent group, and they’re not reflecting their own biases here. This is really how we have consistently thought about this problem. It’s actually how I got into this field in the first place.
D. Allan Drummond (00:11:01):
This boundary here is a cell. Cellular stress such as heat shock enters the cell at exactly one place, and it triggers exactly one event from which everything else is downstream, and that is the conversion of a native protein into an unfolded protein. Then there’s a bunch of stuff that happens. This is conversion of a normally functional thing into a toxic thing that needs to be cleaned up. That’s what I mean the sensor is the danger. I’m going to say this a slightly different way. In a little bit more diagrammatic way, you can see what role this misfolding is playing.
D. Allan Drummond (00:11:43):
So, under normal conditions, you have this heat shock factor 1 bound to DNA. Actually, there’s differences between different systems. You don’t really need to pay attention to this particular detail. It can be bound. It can be unbound, but suffice to say, it’s inactive and held inactive by a molecular chaperone, Hsp70. Meanwhile, a bunch of bystander proteins are floating around in the cell. This is what is thought to happen.
D. Allan Drummond (00:12:10):
When stress hits, that causes the unfolding of these proteins. That exposes a bunch of hydrophobic internal residues that were buried in the folded state. These aggregate together. Titrates away Hsp70, this repressing chaperone. That activates transcription. Transcription is driven, and that transcription drives the production of messenger RNAs that encode a bunch more molecular chaperones that can then tackle these aggregates, disperse them, and assist with the refolding of the proteins. And homeostasis is restored.
D. Allan Drummond (00:12:47):
When I say the sensor is the danger, if you look at this graphic, who’s sensing temperature here? There is a heat shock factor that is not temperature sensitive. There’s a molecular chaperone that is not temperature sensitive. There’s exactly one thing that is actually sensing temperature in this case, and that is this protein that’s undergoing this conversion from a folded state to a misfolded state. This is surprising, or maybe not surprising because we know so much about proteins and limited stability and all these things that this just sort of make sense. You have a heat shock. It produces trash. You deploy the chaperones to clean it up. That’s the paradigm.
D. Allan Drummond (00:13:35):
In a little bit more graphical terms, this is the cellular institute … welcome … with its molecular workers going about their inscrutable business when flames suddenly break out upstairs, and there are screams and howls and people running out into the courtyard, burned horribly, dying, clinging to each other, but it’s too late. Somebody, however, has the presence of mind to pull the fire alarm. Emergency services and/or fire department shows up and says, “Oh, my goodness!” They go about trying to rescue some folks who are shunting them to degradation, and they’re able to eventually restore order and put these folks back to their normal jobs in a disgruntled way.It’s terrible. It’s a bad thing.
D. Allan Drummond (00:14:24):
And why do we think this? Of course, it’s just because we see this every day. This is the conversion of a native protein into an unfolded protein. That’s what heat does to cells and does to proteins. Strangely, after all of this, no endogenous, mature, eukaryotic protein has been shown to misfold in vivo in response to a sublethal heat shock. This sensor has never been identified, and that, for us, was the big puzzle. The very central actor that’s going to drive this entire process has never been isolated to where you can put it in a test tube, and you can study it, you can do mutations, you can shut off the heat shock. All of those aspects that you would hope for instead hadn’t been.
D. Allan Drummond (00:15:09):
And I’ll talk a little bit about the sociology of why that is, but once you believe that the essence of it is, once you believe that this is a misfolding event, well, it doesn’t matter which misfolded protein you use. Who cares about the endogenous substrates? Just use the easiest one you possibly can like firefly luciferase was a fantastic activity assay for something like this. Now, I used this to show the same kind of slide years ago, and I couldn’t say much of anything else, but now I can say that my lab and others–I think most prominently Simon Alberti’s lab–have shown that many of the proteins in cells actually undergo heat-triggered phase separation and condensation responses.
D. Allan Drummond (00:15:49):
The endogenous proteins do this, and the closer we look, the less they conform to the idea that this is really any sort of protein quality control catastrophe for cells. In fact, they know precisely what they’re doing, and they regulate it very, very tightly. So, where this started was with Edward Wallace, where we just asked the question, Edward asked the question using stable isotope labeling methods in cells to track proteins across a stress response and ask what proteins actually just form aggregates in cells. We didn’t have a parts list even for the set of things that we should be looking at.
D. Allan Drummond (00:16:30):
This is just a snapshot of the kind of data that he got. This is the proportion in the supernatant after heat shock, the proportion in the supernatant before heat shock. And the idea here is that in a two-phase lysis of cells, you have soluble proteins that are floating around that end up in the supernatant after you spin stuff down, and big stuff ends up in the pellet. So, we’re just looking for things that start in the supernatant and end up in the pellet. Those are the things that form aggregates of some sort or at least bind to aggregates. There’s a bunch of caveats to this, but every time we’ve gone and looked, either somebody is aggregating themselves or is bound to somebody who’s aggregating.
D. Allan Drummond (00:17:12):
What he found was that there is quite a bit of action going on even after four minutes at this sublethal heat shock temperature. Tons of proteins including every, at that time, known component of stress granules was falling off the axis to some degree except for this guy mildly. But this was only four minutes in, and we went up to eight minutes, as long as eight minutes of this stress. These things happen very quickly. And that there’s a huge number of categories of things like half of the nucleolar proteins crisp up in place. We’re able to show by microscopy. They don’t go anywhere. They form larger clusters than they’re existing in under non-heat shock conditions.
D. Allan Drummond (00:17:50):
He also showed using stable isotope methods that these are entirely reversible. This is completely unexpected, that all of them are restored back to solubility during recovery, and, in some cases, there are eukaryote-specific domains that are different from the catalytic domains that the enzymes are attached to that mediate the formation of these large clusters. So, all of this just does not conform at all to the idea that there is some sort of misfolding catastrophe happening, but we didn’t know much more about it.
D. Allan Drummond (00:18:17):
And just to get into one extra layer of detail, there are specific classes of proteins where across time at 46 degrees Celsius or across a temperature of an eight-minute heat shock, a proportion of the supernatant is either say, unchanged–these are glycolytic enzymes, unchanged here completely insoluble–these are membrane proteins, or change their solubility massively–so components of stress granules in purple here, or things we call super aggregators that aggregate much more quickly than any other category of proteins we could find. Many of them are nuclear proteins or nucleolar proteins, including specific proteins. So poly(A)-binding protein is the classic marker of stress granules, and it nicely forms these little clumps in response to stress.
D. Allan Drummond (00:19:07):
So, this raised the question, is this just a cellular response, or is this somehow a protein-autonomous response? Just a little side note here, many of you may know that last year the Nobel Prize was given in Physiology or Medicine, half of it to David Julius for the discovery of the TRP channels, which are the temperature-sensing channels–or actually four of the temperature-sensing channels–for the discovery of those channels. And I haven’t mentioned those yet. Why are those are the things that are supposed to be doing this work? You’ll notice they don’t appear anywhere in the mechanism that I’ve just articulated, and turns out they’re completely dispensable for the heat shock response that cells do.
D. Allan Drummond (00:19:54):
Amazingly, he got the prize for the discovery of the channels, the isolation and study of them, and so on. We still don’t know how they work. We don’t even know what part of the protein senses temperature. Mechanistically we have almost no idea. There are a bunch of hypotheses. There’s very detailed, like extraordinarily detailed thinking about this, but there is no established model for how temperature is then transduced into the opening of the channel. It’s hypersensitive. 200-fold changes in ion connectivity over a 10-degree range. So, that’s sort of interesting.
D. Allan Drummond (00:20:36):
The other bit that’s I think interesting about temperature just as an object of study is that it’s very unlike a lot of the experiences or a bunch of the signals that cells receive in the sense that at the cellular length scale, the entire cell finds out about temperature at the same time. So in principle, any molecule in the cell can be made temperature-responsive. It’s not like you have to have a dedicated receptor somewhere on the outside of the cell, but then it is involved in some cascade. Instead, at some level, evolution faces the decision, which molecule should be made increasingly temperature-responsive?That raises the question for all of these. We’ve identified 177 different proteins that form these aggregates in response to heat shock. That’s a lot.
D. Allan Drummond (00:21:25):
Presumably, they’re not all talking to one another. In fact, we’ve now demonstrated that. How many of them are listening to temperature directly and transducing that temperature difference into a condensation state? So, Josh Riback, and actually before him an undergraduate student, Alexander Rojeck, now a doctor, isolated poly(A)-binding protein. So purified it recombinantly, put it in a dish, exposed poly(A)-binding protein, the classic stress-granule marker and binder of poly(A) tails and eukaryotic messenger RNAs, put the purified protein in a dish under physiological conditions–and back in 2015 this was, I have to say, a little bit innovative, and found that poly(A)-binding protein itself autonomously with no RNA, without anything else, will condense.
D. Allan Drummond (00:22:22):
This is another case where, at the time, liquid-liquid phase separation was sort of the only thing that was in conversation, and we could show that there is probably a transient liquid state. But by the time you actually see even these beautiful little round droplets, they are gelled. They look utterly unlike the denatured aggregated state of the protein. So, if you acid denature poly(A)-binding protein, it looks like snot, like all the aggregates we’d looked at. We knew we were looking at something else.
D. Allan Drummond (00:22:53):
And it really wasn’t until we made these observations in 2012 right when people were starting to figure out the paradigm that we would then realize that we needed, that we would be a part of. We had a way to understand what we were observing. So, even these things that look a little bit like, oh, they’re sort of gross clumps, if you zoom in them, they’re all little diffusion cluster-associated spheres of gelled protein.
D. Allan Drummond (00:23:23):
So, poly(A)-binding protein autonomously senses temperature, and then Josh was able to show that this was actually formed in a nice phase boundary right across the physiological range of both intracellular pH and temperature, where on one side poly(A)-binding protein is soluble; it can go about its business. On the other side, it’s demixed into these condensed clusters.Then with Chris Katanski, Chris basically did the cell biology side that made it clear that what Josh had discovered here was extremely meaningful.
D. Allan Drummond (00:23:58):
Chris took mutations that Josh had made in poly(A)-binding protein’s intrinsically disordered region, which it turns out is dispensable for phase separation and condensation, but is a modulator of that process. So, Josh could move the phase boundary around in temperature and pH space. Chris could take those mutants, put them in cells, show that he could recapitulate the fraction of pelleted material and cells was directly corresponding to the demixing temperature of the proteins in a test tube completely free of the rest of the cells, and that mutations that disrupted the ability to condense were strains that couldn’t grow properly at high temperature in a single copy at the chromosomal locus.
D. Allan Drummond (00:24:47):
This is one of the earliest phenotypes for a phase separation process. So, disrupt phase separation, you can tell now that the cells can’t grow properly. They have lower fitness; they will die over evolutionary time, and it’s clearly not what’s happening with the wild-type variants. We still actually don’t know what the cellular basis of this phenotype is. There’s tons more to do, but we know a really important thing, which is that in the heat shock response, the formation of these aggregates is supposed to be toxic. And here we’re showing the suppression of the “toxic species” actually makes the cell sad, not happy, doesn’t make it more able to grow. It makes it less able to grow. These condensates are not proteotoxic. They are adaptive, and that is a huge shift.
D. Allan Drummond (00:25:41):
That’s sort of this side of things which is saying, “Okay, most of what we know or thought we knew about the formation of these clusters or condensates in response to heat shock turns out to be incorrect.” And the more we discover, the more we realize that these are incredibly fun objects to study. But what about this process, the chaperone dispersal process? Okay, so Haneul Yoo in the lab–I’ll take you through a little bit of the biology, but I promise you, you don’t really need to remember these names to appreciate her results–she tackled the question of in vitro dispersal of poly(A)-binding protein condensates back to functional monomers by molecular chaperones that are induced in response to heat shock.
D. Allan Drummond (00:26:28):
So, these things have been studied for a really long time. Their biochemistry is relatively well worked out. We knew exactly what system we wanted to challenge with the dispersal of poly(A)-binding protein condensates, and that was this molecular disaggregation system, which in budding yeast, in almost everything except for animals, consists of Hsp104, a big AAA plus ATPase. There’s now I think, pretty good circumstantial evidence that VCP plays this role in the cytosol of mammalian cells. Proteins called Hsp40, chaperones called Hsp40 and Hsp70, which we had mentioned before. So, these three chaperones can collaborate to disperse. They’re known to be able to collaborate to disperse misfolded protein aggregates.
D. Allan Drummond (00:27:15):
And how do they do it? Well, imagine there’s some big aggregate. The Hsp40s initially bind in the substrate. They recruit Hsp70, which uses ATP hydrolysis to get itself into a high-affinity state. Those Hsp70s, if they’re multiply bound, attract Hsp104, which then is this giant motor protein that’s activated by the presence of Hsp70. It drags the aggregated protein through a central pore and noodles it so that it then can be bound by chaperones to protect it from aggregation. It’ll either re-aggregate or spontaneously refold. So, you do this enough times and hopefully, everything disaggregates and refolds.
D. Allan Drummond (00:28:01):
Now, that’s hopefully, but really the hope is rarely realized. These are Haneul’s data, but these look like everybody else’s data in the field. It was just important for us to convince ourselves that we weren’t missing anything. So, this is the activity of the molecular chaperones, those same molecular chaperones I just told you, against firefly luciferase aggregates. Why firefly luciferase? Because it has a fantastic activity assay. It produces light with really, really low background, and it’s a misfolded protein, so it turns out to be a thermal labile protein, so you can misfold it easily with temperature. If you do that, then hey, it’s a misfolded protein. All misfolded proteins are basically the same. Right?
D. Allan Drummond (00:28:53):
So, put misfolded luciferase in a test tube, add molecular chaperones and ATP, and see what kind of dispersal you get. There’s a few different conditions here. This is basically Hsp104 and Hsp70, a nucleotide exchange factor for Hsp70, a couple of Hsp40s either working independently or synergistically. It turns out synergistically. This is a discovery from Bernd Bukau’s lab. Synergistically, they work better together, and they can disperse up to, I don’t know, in our hands 40%, in other people’s hands 60%, something like this. After two hours, they can restore that much activity for luciferase.
D. Allan Drummond (00:29:34):
Now, dirty little secret, this is at a chaperone substrate ratio of about 40 to one. Now, I am a professor of biochemistry who’s never taken a class in biochemistry, but I’m pretty sure the enzymes are supposed to not be consumed in the reaction, and you shouldn’t need this. Why is it so high, and why are they so inefficient? This is terrible if this is the job that these molecular chaperones are supposed to be doing. One possibility is it’s the wrong substrate.
D. Allan Drummond (00:30:09):
So, what Haneul did was take this putitive, endogenous substrate, not foreign substrate, but endogenous substrate, work out solution conditions where she could use fluorescence anisotropy to monitor the dispersal of poly(A)-binding protein back to functional monitors. The essential principle is that she labels RNA, and that will tumble until it’s bound by poly(A)-binding protein, at which point, it’ll slow its tumbling, anisotropy will increase, and she can read that out using a graph that looks like this: time versus change in anisotropy. Here’s the no ATP case where the molecular chaperones are there but they can’t do anything because they don’t have ATP. And here’s the plus ATP case. She actually made us a biochemistry lab finally by doing the plus-minus ATP experiment for the first time.You can see there is pretty rapid dispersal of these things and pretty fast.
D. Allan Drummond (00:31:13):
But now, let’s do the heads-up comparison of the two substrates, same concentrations, same concentrations of chaperones, as identical as we can make them. So, these are those results, and we published this earlier this year. It may seem like there’s an error here. Here’s poly(A)-binding protein popping out of a solution back to functional monomers within 20 minutes or so, back to nearly 100%, and luciferase is just doing nothing. It’s not doing nothing. You just need to use a log scale to see the activity. This is the same substrate as before, but now Haneul’s dropped down the concentrations in some cases below stoichiometric or two as much as five fivefold excess of things, but not 40 fold.
D. Allan Drummond (00:32:03):
And it turns out the poly(A)-binding protein’s delighted to do things under those circumstances and firefly luciferase just sits there. Now, remember, firefly luciferase is supposed to be a model protein, a model for the endogenous proteins which had never been studied. If you read our paper, it is just a kind of slog through all the different ways in which luciferase and other proteins that have been studied in this way are not good models for the endogenous substrates themselves.
D. Allan Drummond (00:32:34):
It’s a different molecular problem that’s being solved. Basically, for those of you who really want to think in detail about this, it is the lack of the back reaction. There is no recondensation reaction. There is a reaggregation reaction that the chaperons have to fight, but there is no recondensation reaction that needs to be fought by these molecular chaperons, so they’re insanely more efficient. This is two orders of magnitude faster for dispersal of condensates than misfolded proteins.
D. Allan Drummond (00:33:00):
Okay. So, this old model of stress and trash and cleanup is now a … This is a signal. This is a response, and the chaperones are doing regulation. I’m getting close to time here, so I just wanted to mention-
Bede Portz (00:33:20):
Allan, may I interrupt you-
D. Allan Drummond (00:33:21):
Bede Portz (00:33:21):
… to ask a question about.
D. Allan Drummond (00:33:21):
Bede Portz (00:33:22):
… what you just said, which struck me as pretty profound. The model in papers is so often that condensates are en route to some aggregated bad outcome. But what you’re saying here is that they might be an intermediate en route back to monodispersity from an aggregate. Am I correct in that?
D. Allan Drummond (00:33:46):
Absolutely right, yeah. Yeah, yeah. I think there’s a really fun conversation to be had about all the things that condensates can do because I think there was an early notion that they must be compartments for doing valuable chemical reactions. And the more you think about it, the more you’re like, “Well, that’s just one tiny fragment of the edge of the veneer on the surface of what condensates can actually do.” When your molecular operation’s essentially, can you push things together or pull them apart? Just start thinking what you can do there.
D. Allan Drummond (00:34:16):
I just want to point out here that the activity that we are really interested in, the deep sense in which the heat shock response is the condensation response, is that the cell is using a phase transition to sense temperature in the first place, that this is actually the way that it does it. If you think about it, temperature sensing, it’s really hard. Why do we not know how the TRP channels work? It’s because it’s a really hard problem to figure out how you get 100 kilocalories per mole of enthalpy change in a single molecule, but you don’t have to. That’s the point.
D. Allan Drummond (00:34:51):
For multiple molecules collaborating with each other in this highly cooperative way, suddenly, with phase transitions, it’s easy to discriminate a one-degree temperature difference or a fraction of a degree. This is exactly what phase transitions are good for. So, the first thing is that, yeah, we absolutely think that these molecules are doing things and that maybe the condensates themselves are byproducts. We published a review in 2019 saying it’s not just the products, it’s the process of condensation that’s important.
D. Allan Drummond (00:35:24):
Sometimes you may see these poly(A)-binding protein contents. They don’t seem to contain anything. They don’t seem to have any sort of activity, but they are the result of the crucial activity of sensing temperature by poly(A)-binding protein. And that then is going to suppress any soluble activity the poly(A)-binding protein has, so this is the sequestration or inhibition model. You can imagine inverse cases in which the monomer is inactive and only the condensate is active or even more highly active when co-localized with substrates, things like this, and that the chaperones are there regulating this process back and forth.
D. Allan Drummond (00:35:57):
So you decide where you put the function, and the chaperones will say, “Great, I will help you get out of one phase. I’ll actually help you get into another phase.” So, there are also another set of molecular chaperones and small heat shock proteins that are thought of now as sequestrases or aggregases, so they promote the formation of these things essentially by acting as nucleators. I think this is going to be an incredibly rich area, but it is a way to reframe the aggregation and clean up into a, well, actually we’re just pushing molecules together and pulling them apart, and now let’s really think about all the activities that can be executed in that way.
D. Allan Drummond (00:36:32):
Does that answer your question, Bede?
Bede Portz (00:36:34):
D. Allan Drummond (00:36:35):
Cool. And thank you for that. Okay, so if this is the view before, squint your eyes and ask yourself what part of this model can we preserve? What should we preserve? What part were we really making up that we didn’t know but we thought that we knew? When I say squint your eyes, I really mean if you squint enough, you realize you can’t really see the expressions on the faces of the molecules. We didn’t really know that. There’s another activity that looks like this. Everybody’s happy, sort of. It’s a fire drill. We know exactly what they’re doing. There is an increased endeavor.
D. Allan Drummond (00:37:15):
There actually is a fire. Fine. The nascent polypeptide coming off the ribosome is sensitive to temperature in some way, maybe the undergrad up here. But everybody else, just go outside, go to your designated assembly locations. Everybody hold hands, continue to work on your mobile devices. Somebody pulls the fire alarm. The fire department shows up. Thank you, everybody, for following your detailed emergency plan. Thank you for assembling, everybody. We’ve taken care of the bit of business. We’ve cleaned up the very small sensitive part of this whole process, and they’re going to be fine. Hopefully, maybe not, but everybody else, okay, now, resume your work. Resume whatever it is that you were doing. Go back inside. Oh, can you go back? Oh, you can’t go back inside just yet because you’re all holding hands. Okay, so now just let go of each other and go back inside the building. Regulated and intentional.
D. Allan Drummond (00:38:13):
So, one way to think about this is that if we revisit that graph that I showed you right at the beginning, that the stress condition is going to take these soluble molecules, and in a wide range of different stress conditions, promote the formation of different kinds of condensates and different distributions of condensates. We have 177 different proteins doing this in all sorts of different cellular compartments all at the same time just during heat shock. Now, imagine what kind of richness we can get if we start doing this over a range of other stress conditions, which we are in fact doing right now, focusing mostly on the RNA components, which I haven’t even mentioned today.
D. Allan Drummond (00:38:51):
All of those can recruit molecular chaperones that are inhibiting either HSF1 or some other relevant set of transcription factors. This is a completely reusable motif. But then those same factors are induced heavily during stress until they rise to a concentration where they’re capable of dispersing these condensates. Then they’re in excess, and so in excess enough to come back and silence the response that they started in the first place. Sensation, regulation, return of homeostasis, with an appropriate response executed in the middle. This is the way that I think that we can productively think of the heat shock response as a condensation response intrinsically from sensation all the way through that restoration of homeostasis.
D. Allan Drummond (00:39:40):
Why think about that? It’s really a statement that we should be thinking about temperature as a signal, not a stress. And we know that this happens. There’s a bunch of fungi, if you care about fungi, which I happen to. There’s a bunch of fungi that sit out in the environment, the dimorphic fungi. They detect that they’ve arrived at their vertebrate host by an increase in temperature rather than do a developmental transition right into their yeast forms. Candida albicans does the exact opposite thing when it breaks our skin. It senses the increase in temperature, and then it moves from a yeast form to a hyphae form.
D. Allan Drummond (00:40:13):
And we’ve argued that Saccharomyces cerevisiae, the primary physiological heat shock that it experiences is, it’s sitting on a grape eating glucose, it needs planetary dispersal because it can’t form fruiting bodies, it needs animals to do it. It gets eaten by a bird. Birds are 42 degrees Celsius on the inside. It knows precisely what is happening to it at that time. It is not stressed out. It’s just preparing for planetary domination, which it’s probably going to be necessary to go through some spore form in order to do that.
D. Allan Drummond (00:40:41):
And of course, this is not just fungi, it’s in the immune system all over the place. Fever is now well understood too. 5% of the time fever is restricting pathogen growth like you might have thought, certainly, like I thought as a kid, but most of the time, it’s just serving as a red alert for the immune system. So, neutrophils are circulating immune cells. They get a local pH signal and a global temperature signal, and they suppress apoptosis right in the places that they’re needed to proliferate to fight infection. And this happens in T cells. This happens in B cells. It happens all over the place. It’s just a generalized signal.
D. Allan Drummond (00:41:15):
We know almost nothing about this. We know almost nothing. We don’t know how the temperature is sensed. We don’t know the intracellular changes that happen. We know almost nothing about the condensation landscape that’s happening inside these cells. This is something that is newer that we’ve just started up in my lab right now, but we need 1,000 people working on these sorts of problems because they’ve laid dormant because essentially the idea was, well, it’s all just a giant stress response, so it’s probably all the same. I think that’s just demonstrably not the case. Okay, so I will finish there. Thank you all for listening in, and I’d love to take questions.
Jill Bouchard (00:41:59):
Awesome. That was the coolest analogy ever.
Bede Portz (00:42:06):
Kamran Rizzolo (00:42:11):
Hi, Allan. Thank you very much for the talk. I just had a question in regards to using yeast as the model here, and if we can take a step back and think about the story that you’re telling us in the case of humans. I was wondering if you have looked at the role of the concentration of the proteins inside of the cell because, as you know, in yeast, concentration of proteins goes up a lot depending on the cycle it’s in. So, for example, in quiescence, you get many, many proteins doing the things that you’re describing. So, I was just wondering if you had a chance to look at that aspect.
D. Allan Drummond (00:43:04):
We haven’t, and I can say I don’t think that’s a particularly good choice. We have to make hard choices. I think it’s a really interesting question about the role that protein concentration is going to play. We do think a lot about the cellular abundance of proteins, and I can just say that it’s doubtless going to play a role. But one of the fundamental questions that we have been worried about is to what extent … Take stress granules. They are composed of dozens of different proteins. There are two extreme models. One is that all the individual components autonomously condense on their own, dictated by their own intracellular concentrations and so on, and then are brought together by cellular processes, like cytoskeletal transport processes.
D. Allan Drummond (00:43:56):
And the other is that they all actually co-condense together, and they require each other, at which point, this effective concentration is really summing over all of the individual species. Nobody knows which one it is. Every time we’ve checked, when we’ve taken proteins out and purified them, we find that it is autonomous. So, it suggests to us that probably the concentrations of these things is going to be important. The individual concentrations of the individual molecular species, not the collective concentrations, that is going to be important. But we’re also guided by the fact that it’s just much easier to express and purify highly abundant proteins, which is one of the reasons we started with poly(A)-binding proteins.
D. Allan Drummond (00:44:36):
So, the low abundance proteins are way less studied, and I think there are really interesting questions about the degree to which they can undergo these same sorts of things or piggyback on other systems that get them to a high enough concentration, local concentration, that they can execute these sorts of activities. I’m not certain that answers your question, but basically it’s like, “No way. I haven’t really looked and definitely somebody should.”
Kamran Rizzolo (00:45:01):
Yeah, thank you so much. Yeah, thank you.
D. Allan Drummond (00:45:04):
Of course. Anyone else?
Carlos Castañeda (00:45:11):
I have one question. Oh, sorry.
D. Allan Drummond (00:45:16):
Carlos Castañeda (00:45:17):
Should I go here? Should I go ahead?
Bede Portz (00:45:18):
Go ahead, Carlos, and then we’ll do Jian Guo.
Carlos Castañeda (00:45:23):
I love this model. I’ll be biased and say that because I think a lot about the chaperone effects and thinking about what they’re doing. I think this question about what is the sensor for a heat-based response that that’s something that I think has intrigued me for quite a long time, so this is actually really cool. I had a question about thinking about local temperature effects and cells. Translating this stuff to humans where I know we have our regulated temperature, but as you were kind of mentioning towards the end, there are instances of fever that we have when we get sick. What is known … because maybe I should know this, but I don’t … about local temperature effects within a cell? You kind of tell …
D. Allan Drummond (00:46:11):
Yeah, so I think there’s two things to say. One, as someone who knows a little bit about skin, you should definitely be thinking that our encounters with temperature are not limited to homeostatic mechanisms. We are constantly reaching out and feeling warm things and moving ourselves around sensing temperature in ways that preserve our skin barrier and other parts of our epithelium from damage by temperature. A lot of, I think, the heat shock response, we’re just eating hot food and our tongue gets burned occasionally. It’s pain, but that’s a physiological thing that our cells have learned how to adapt to very, very rapidly. And we have all sorts of sensory mechanisms but also the cellular mechanisms that can do recovery from a wide range of thermal perturbations and that route. Those things are largely unstudied. People are so fascinated with homeostasis that they forget and just touch hot stuff all the time.
D. Allan Drummond (00:47:11):
Now, I know that you’re not asking that question. You’re asking this question about the variation of temperature within a cell. And I have to tell you when I hear this, I hear the, oh, mitochondria run at 50 degrees Celsius and those sorts of things. I have to say I don’t believe it yet, just I am not ready to actually admit there is a problem to be solved there just yet. It’s clear that, of course, that you can have these sort of temperature variations, but the intuitive thinking to the level that my thermodynamics goes tells me that those gradients disperse incredibly rapidly and that the measurements that have been done claiming otherwise are really not reliable in my mind. I’d like to see them a few more ways. That can just be my natural conservatism, but my approach to this.
Carlos Castañeda (00:48:02):
That’s helpful, thanks.
D. Allan Drummond (00:48:04):
Bede Portz (00:48:05):
Okay, Jian Guo here at Dewpoint.
Jian Guo (00:48:08):
Okay, thanks Allan for this very interesting and wonderful talk. Actually I love it. I have a question. Recently, people found that the neutrophils can be used as a predictive marker for tumor relapse or resistance. Actually, I found that your last slide is very interesting. I’m not quite sure about this one, but based on my … What I’m saying, let’s say you see the neutrophils can suppress apoptosis based on the pH or temperature. However, I don’t know. Can you elaborate a little bit about the mechanism? How does this happen? I’m really interested about this one. Yeah, thank you.
D. Allan Drummond (00:48:51):
Yeah, yeah, thanks for the question. No, I have no idea. First of all, I should point out that those are not our results, but the little bit of background here is that yes, it is known that neutrophils respond to both the local pH signal, local acidification or acidosis as a sign of in this particular area, something is going wrong. The tissue integrity has been breached. There’s an actual infection. Tumors also have this local acidosis, ischemic shocks, and other things. And then you get that second hit of temperature, and you get a synergistic suppression of apoptosis.
D. Allan Drummond (00:49:27):
For those of you who don’t know, neutrophils are just characterized by just killing themselves all the time. So, suppression of apoptosis is basically the you are now needed. Stop killing yourself. We need you at this time. So, that’s a major activity that they undergo in order to proliferate and start to do their actual work. How that happens, we’ve no idea. We just have no idea. Now, the place where we get excited, where my lab gets excited is in 2020, my graduate student, Cat Triandafillou, published a paper in eLife that showed that transient intracellular acidification, so a drop in pH in the cell that always follows heat shock, is necessary for the induction of the heat shock messages under very specific conditions.
D. Allan Drummond (00:50:20):
So, it’s been known forever that this sort of acidification happens, but it had been thought to be some pathological effect. I think increasingly we’re realizing that this is really a part of the physiology. The cell uses both the pH as a kind of second messenger, the protons as second messengers, and temperature or some other set of stresses to integrate, to tell themselves what’s going to happen. Now, mechanistically speaking, I just want to refer you back to the phase diagram that I showed. I don’t know if you have it in your mind, and I’m not sure that I can pull it up fast enough to not waste everybody’s time, but the phase diagram was temperature and pH and a line curving through the middle of those things.
D. Allan Drummond (00:50:59):
So, the individual poly(A)-binding protein molecules are able to do this integration of pH and temperature themselves and decide condense on one side, don’t condense on the other side. That provides this really interesting mechanism, and Simon Alberti’s group has also shown tons of pH-responsive and temperature-responsive. I mean Ded1 there is also this sort of thing. This example is repeated all over the place. So, the place where we’re looking in neutrophils is look at the condensates. That would be the obvious way to integrate this signal that would follow precedents that have now been established in other systems. But it’s all completely yet to be discovered.
Bede Portz (00:51:40):
There’s a related question in the chat from Albert Calzada. Do you want to unmute and ask your question? If not, I’ll ask that question because it was also my question. You started your talk with this heat shock elicits a response that overlaps with the litany of other so-called stresses, and then you focused on heat shock. So, what are the sensors of and triggers of those sensors for other stresses? And Albert cites hypoxia, but you could apply any of them. Is PAB1 universal, or are there PAB1-like things for other stresses that have been encountered evolutionarily?
D. Allan Drummond (00:52:29):
Yeah, so it’s a great question. I think in many of the stresses that have been well studied, there are specific sensors. For nutrient depletion you get the mTOR pathway, essentially there ends up being for Gcn2, there ends up being sensing of uncharged tRNAs as a sign that the cell has poor levels of amino acids. In hypoxia, it’s HIF-1-alpha, so the sensing mechanisms there have been worked out. Then the question is, how do they then converge on a common pathway that ends up with their recruitment of all of these molecules into the same kinds of structures? We don’t know if they’re really the same. It’s still actually we’re at the point in the field where people are still putting together the parts list.
D. Allan Drummond (00:53:15):
Roy Parker’s group has published this observation that no matter what stress you give, you get the same messenger RNAs in the stress granules. I have to say we don’t think that’s that that’s really the case. I don’t think that’s the end of it, but it’s just to say that there’s a bunch of different things that are yet to be determined even. So, what could be the convergence pathway? And here, I get to just happily pay off the little nugget that we just litigated in the previous question, which is you just do it through pH would be a good way to do it.
D. Allan Drummond (00:53:44):
So, Simon’s group, I think, was really the first to propose this, that there would be a second messenger type of effect from pH. And it turns out that pH also accompanies heat shock and accompanies all these other stresses. You get this transient drop in pH. Whether that’s sufficient signal, I don’t think so, but it’s clear that there exists pHs that will cause poly(A)-binding protein to condense at normal temperatures. So, we know that that is already possible in the case of at least one protein that we’ve studied, and time will tell whether or not that’s really sufficient to recapitulate the global response. But that’s my bet is it’s going to be pH or some other similar or small molecule type of effect that follows those stresses.
Bede Portz (00:54:32):
There’s one more question in the chat from Pritam. Do you want to unmute and ask your question?
So, the question is, if these condensates or aggregates are adaptive, why do we need chaperones to disaggregate them?
D. Allan Drummond (00:54:58):
It’s a great question. It’s a perfect example of the change in thinking that I think is required. They’re not aggregates. They’re just condensates. So, then the question is why do condensates need to be regulated? And one of the answers is, well, you could either try to program the individual condensed molecules to just know exactly what condensation formed to be in at all times. That’s formally possible, but there are very specific intellectual advantages to disentangling the condensation response itself from the dispersal process.
D. Allan Drummond (00:55:36):
In particular, in the case of the stress response, the question is what do you want? What’s the goal? The goal here is to sense and appropriately respond to this new signal that’s arrived. So, it doesn’t matter whether you think of it as a stress response, doesn’t matter whether you think of it as it’s a signal. I’ve been eaten by a bird help. I need to induce my spoilation program. You don’t actually care about detecting the end of the stress. You care about detecting the end of the stress response.
D. Allan Drummond (00:56:04):
So, the onset of the signal means you condense. Then you execute this program that involves inducing molecular chaperones along with a ton of other things. It’s always more than one regulon. In the case of heat shock, it’s HSF1 is the thing that we talked mostly about, but also Msn2/4, which is this other regulon that does all sorts of membrane remodeling, ceramide biosynthesis, and all this other stuff, the stuff that is actually required. The HSF1 regulon is just the thing that’s going to produce the regulators of the condensates. So, those things then build up to a certain level that is sufficient to disperse the condensates. The cell then knows once the condensates are dispersed, it knows that it has produced enough protein. It has provably completed the response that it had intended to do.
D. Allan Drummond (00:56:52):
Now, remember what the alternative is is you have some autonomous condensate. It just goes in when the temperature is high, and it comes out when the temperature is cold. There, you would defeat that entire signal. As soon as the heat shock had dispersed, you would forget that you ever saw that signal in the first place. So, in some sense, the condensation is the locking-in of a memory of we now need to do something, and the chaperones are the way to disentangle the, okay, now, it’s time finally to forget. We’ve done all the things that we need to do. We can put that into our past. We have remodeled ourselves for whatever new condition that we need to do.
D. Allan Drummond (00:57:32):
Here, it is crucial to think of, it’s not that the chaperones are cleaning up the condensates. They’re just acting as the regulators. As important as condensation is, decondensation is the other activity that is equally biologically important.
Bede Portz (00:57:46):
One last question from Adam Klosin in the chat. Do you want to unmute and ask your question, Adam?
Adam Klosin (00:57:52):
Sure. Thank you for the great talk, Allan. That was really, really informative and really fun. I was wondering if you could comment about the differences between how cytoplasmic proteins respond to stress and nuclear proteins because what you talked about was mainly about PAB1 and how it responds to stress and how the condensates form in the cytoplasm, cytoplasmic condensates. But you also mentioned HSF1, which is the protein that forms bodies in the nucleus. But they seem to be like a secondary response, right, or do you think they’re also directly triggered by stress? Could you just comment on that the difference that there might be between cytoplasm and nucleus and stress response and how you are thinking about that?
D. Allan Drummond (00:58:45):
For sure. Thanks for the question, and it’s a great one and, I think, a deep one. I think the first thing to say is that, yes, these sorts of responses are happening everywhere. Even in the initial dataset where we collected things, I think it may have passed by too fast, but the vast majority of the super-fast condensing proteins that we identified were actually nuclear or nucleolar, not cytoplasmic. In fact, there’s only one exception. There’s only one of those that’s cytoplasmic. Almost all of them are sitting around in the nucleus.
D. Allan Drummond (00:59:15):
When we tag those things, they look like any other thing that you’d see. There’s little spots in the nucleus instead of little spots in the cytoplasm, so it’s not like they are immediately intrinsically different, except for the nucleolar protein. Here’s a fun thing. It’s so hard to see. Yeast is already small. The nucleolus is this little crescent instead of big blobs. How do you ever see the condensates form? Turns out we had to wait for late anaphase, and then the nucleus is peeled apart because yeast does closed mitosis, or it doesn’t dissolve the nuclear membrane. The nucleolus spans the little gap between the daughter cell and the mother cell, and you can see these little spots along the little extruded nucleolus. So, they’re happening all over the place.
D. Allan Drummond (00:59:57):
Then the question is… they’re happening, of course, in chromatin. Here, I just want to guide you immediately to my colleague David Pincus at the University of Chicago who’s studying these things pretty intensely. There’s been a ton of other work on the formation of HSF1 condensates themselves. These are chromatin-remodeling events that seem to be tightly coupled to the execution of the response itself. Yeah, so we think they’re going to be functional. They’re going to not involve kind of soluble proteins in the usual way, but rather co-localization of regulons.
D. Allan Drummond (01:00:32):
My sense … I guess this gets to be my time to opine on these things, but it still seems to me to be very early days and a lot more heat and light in the transcriptional kind of world. I’m still waiting for the real reliable paradigms and model systems to emerge where we can just sort of say, “Okay, this is exactly how it works,” and everybody’s not arguing with one another for priority, but it’s so clear that the phenomena happening. That’s obvious, and it’s up to us to get the right models and methods and so on to really figure out what’s happening.
Adam Klosin (01:01:10):
Cool, thanks a lot.
D. Allan Drummond (01:01:11):
Bede Portz (01:01:12):
All right. Thank you very much, Allan, for a very stimulating talk and discussion, and-
D. Allan Drummond (01:01:20):
Thank you all.
Bede Portz (01:01:21):
… thank you so much for joining us.
Jill Bouchard (01:01:22):
Yeah, you guys…
D. Allan Drummond (01:01:24):
Jill Bouchard (01:01:25):
…made a very nice day for us. Thank you so much, Allan, for joining us. Thanks all for joining via Zoom and at the kitchen table, and we’ll let you know when we have another one.
D. Allan Drummond (01:01:36):
Jill Bouchard (01:01:39):
Thanks again everyone.
D. Allan Drummond (01:01:39):
Question from David Dachille: One thing you mentioned in your talk was that at different temperatures, the distribution of different condensates will vary. Does this imply an optimal temperature for each condensate mediated function? Does the proportion of a condensate within the condensate population dictate a function? And lastly, is there an optimal temperature for each condensate mediated function?
It was a really interesting talk got really got me thinking about condensate mechanisms!
Allan’s Response: It’s an interesting question. We think — based on results showing that the same protein condenses at different temperatures in organisms adapted to different thermal niches — that condensation does not have an “optimal temperature” so much as that individual proteins have evolved to condense at a temperature which has the same evolutionary significance. E.g. heat shock for a cryophile might be 32C, whereas for a thermophile it might be 50C, and homologous proteins in each organism would condense at the respective temperatures. Condensation, like so many other molecular properties, is an evolvable trait. Consequently, functions associated with condensation are not tied to any particular temperature. To the questions of whether the proportion of a condensate within a population dictates function, if I understand correctly, generally if condensation itself has a functional meaning (not guaranteed in every case!) then one would expect a relationship between proportion condensed and function — though this could be more function, less function, or some non-monotonic relationship! The functional repertoire of condensates is barely explored at this point, despite considerable creative thinking by many folks. Much to do.